Working on a herd of peripherals this weekend


I adjusted the tension of the main card reader belt, but it is still giving off read checks and stacker/hopper rollers seem to stall or roll really slowly. There does not appear to be a clutch in the mechanism, so this is pathological when the main belt is rotating at full speed. I suspected either a loose setscrew on a pulley or some issue with the motor itself stalling. I needed to do a bit more diagnostic work first. Turned out to be a loose setscrew, once adjusted the machine settled into its one and only problem, the sticky solenoid.

Removed the solenoid to look into the situation more closely. The solenoid itself rotates and snaps back perfectly, the issue is with the shaft and bearings for the picker shoe. I spent about an hour fighting to remove the shaft, the first step of which is to remove a circlip and then pulling the shaft out from the bottom. Getting the right lighting itself was a bit of work.I took off both sides clips and inspected the bearings on both sides.

Once off, I discovered that bearings, particularly the top one, were clogged with fine sandlike material and solidified grease. I corrected it by soaking the bearing, cleaning it out and relubing it, then replaced the shoe, shaft, and solenoid. It feels great now.

Now that I put it back together I will have to go through the multiple steps of adjustment necessary before I can fire up the reader for another round of testing. I feel good about the situation, however, as it does appear that this was the issue plaguing the reader operation.

I did a pretty quick set of adjustments, put in the cards, and the pick errors are totally gone! It runs through the deck of cards I have, although there is a set of blank cards with a particular printed legend on them that will experience read checks about every 3 or 4 cards, most likely due to excessive slipping during transport. I tried some really swollen, mistreated cards that came with the punch and these read fine. This may be an adjustment, such as the stop position for the pick shoe, or something in the transport path, but I have a reader that is pretty reliable already.

I have some cards in my home office that I also ran through the machine tomorrow, which narrowed down what characteristics produce the read checks. A  few of my cards experienced read checks too. Oddly, the only cards that get read checks are those with the 45 degree clipped top edge on the right side, whether the original batch I tested or the new ones from my office.  This may be a case of that top right edge exposing light onto a photocell, which the logic may interpret as a card that is prematurely past the read station, although it legitimately has not exited the read station yet. It could also be slipping cards or some other issue.

I may do a more complete set of adjustments to see if I can get it to handle anything including those cards that cause the problems.

Next up is to complete the interface logic that mimics a 2501 card reader while reading from this reader. There are some twists, because the 2501 has a station that a card is delivered to, called the preread station, requiring one cycle of the reader before any actual reading occurs. This may be emptied by a NPRO (non-process run out) button on the 2501 that moves the cards to the stacker. When the hopper empties, there is still one card sitting in the preread section.

If there are more cards to be read by the program when the hopper empties, those cards are placed in the hopper and the Start button pushed again. The reader continues reading with the card in the preread while feeding the next card from the hopper . If this is indeed the last card to be processed in the deck, when the hopper empties, pushing Start again without adding cards will signal a special last card state that reads the final card sitting in the preread station.

The 1130 adapter is very involved with the detailed mechanism of the 2501 reader - there are three sets of contacts that indicate various points during the rotation of the mechanism that constitutes one card movement. The adapter will watch for the photocells to indicate that a card is moving into the read station, then it writes some timing pulses on a magnetic drum inside the reader. These pulses indicate when each card column is under the photocells. At the end of the card, the adapter will erase the pulses on the drum.

My logic will need to deliver signals at the right times that claim to be from the three contacts, accept and acknowledge requests to write or erase the drum, and map errors to the right set of symptom signals for the 2501.


My absolute filter and cleaning supplies all arrived, so I began the restoration and checkout.I have removed the old gasket foam that was crumbling at the outlet of clean air where it would be forced into the disk cartridge. I have replacement material and will glue it in place after I do a thorough blow-out, wipe down and cleaning. Once the air path is clean I will put on the new absolute filter and recreate the missing prefilter.

 I did get power into the unit but suspect that one of the voltage regulators may not be working properly. Also, one of the filter capacitors has zero volts across it in operation, I believe this is the +5V filter. I will begin to track down the status more finely and make repairs if needed, until I have good power levels. Once that is clean, I will reinsert the logic cards into the cage and do some more checkout.

My parts arrived for the interface I will build - the special chips that work on the DEC unibus levels and requirements, plus the terminator resistor packs. I am in the midst of designing the terminator and cable connection cards for home production. I have to work out the routing of signals for the terminator board, then I can assemble that card. The other card for the interface may change depending on what logic I think I should put on that PCB rather than placing it at the other end of the dual ribbon cable bus.

The blower itself has foam between its input and the logic card cage, to pull air through the cage and cool the logic cards. This foam should be replaced too, as it will be flaking sooner or later. I began removing the blower, but need a different tool which is now on order.

I have now verified that all three voltage regulators are working properly and putting out clean +5, +15 and -15V DC. It is therefore time to reinstall the logic cards and see if we can get this to allow 'load' with an unlocked door.

I opened and cleaned the two disk cartridges with pure isopropyl alcohol and kimwipes, not only the platter itself but the insides of the cartridge, then sealed them both in plastic bags to protect them from new contaminants. One disk is in excellent condition, the other has had some head/disk interference on two cylinders but nothing that raised an edge or dug a groove. Further, since the interference did not get dragged across the entire surface by seeks and by power down, I believe it was a transient condition that cleared up after a short period of time. I will be a bit more cautious about introducing the second disk into the drive, but believe it should be okay for use.

I cleaned the heads, which look pristine and had no oxide come off when they were wiped with IPA. This is a good sign of both good maintenance for the drive and lack of any head/disk interference for this drive.


I brought up power on the card punch, to see what condition it appeared to be. The basic transport mechanism seems usable - although it ran slow because it drew a lot of current through my poor extension cord, it did seem to activate solenoids and motors properly with the 'clear' function. I haven't looked closer to see whether the read head and the punch station are functional or if they are somehow frozen or damaged beyond repair. Nothing appears bad on the outside.

This was a key test because I will commit to buy a laser chiller, a unit that will pump cooled glycol and water through the punch head to remove heat, and the unit will report on the temperature of the returned coolant so that I can moderate my use of the punch head based on the thermal situation. I did buy the chiller and will pick it up sometime during the week.


I didn't get to the printer at all this weekend.


I was just not happy with the variable behavior of the keyboard, with every photocell having different start and end resistances, magnified by variations in light intensity from the bank of 10 incandescent lamps aimed down the channels. The voltage dividers weren't always reaching the two thresholds that move the schmitt trigger inputs to the proper new state. I decided to take a more aggressive approach to get the circuit rock solid.

After replacing the small incandescent lamps with a bank of LEDs, I knew I had pretty uniform light in each channel, but also light that would remain much more constant over time, plus the lifetime of LEDs is much better than the old bulbs. I then developed a comparator circuit using an LM339 quad chip and positive feedback to force it to swing crisply. After a bit of experimentation, I found that a 3.7M resistance in series with each photocell produced exactly the results I wanted, except for channel 10 which had the photocell I replaced recently. That new photocell had very different characteristics and worked best with a 200K resistor.

 I decided to keep the keyboard interface board I have, but interpose a new additional board containing three LM339 chips and the 45 resistors, using surface mount devices so that I could try my hand at a home manufactured PCB without killing myself with all the drill holes that would be needed for ordinary hole-through parts. I have designed it for a 3" x 5" board size, which will have quite a few holes in any case, for the 9 inputs, 9 outputs, two power lines, and for some wire jumpers that will span sections that couldn't be mapped purely with a single sided copper board. No arrangement I could imagine would not force some lines to have to cross others, even though I spent hours working on this to reduce those to a near minimum. Again, the drilling is going to be tedious, even though I have a nice high quality drill press in the garage. The PCB plastic is hard, it takes hard bits and some care to make it come out right.

My first try was to use the laser printer in my home office with the 'approved' transparency film. It produced a blotchy black that is likely to result in some broken traces where copper is inproperly removed. It is possible I could touch it up with ink, but going to try an inkjet alternative first.

I discovered that our Canon MX430 printer driver does not support flipping/mirroring of output, but I was able to produce the proper new image from my PCB design software. The reason we want it mirrored is to put the ink flat against the board rather than separated by the layer of transparency film - produces a sharper image on the board.

The printer driver also has no setting for transparencies, only regular and photo papers. The laser transparencies yield very poor results, with the ink pooling on the surface and running. I will get some inkjet printer transparencies and give it a try, but I may need to visit a Fedex-Kinkos or similar to have them produce a high quality transparency that has the solid blacks I need.


I did not get the logic analyzer and Arduino gear set up this weekend, but this will be a priority activity during the week.


The paper tape reader board will be home made, thus it is backed up behind the keyboard and disk boards I am preparing. I did a bit of design on the PCB but there is still quite a bit to do before it is all routed and ready to build.


I did not get to the paper tape punch this weekend, either.

No comments:

Post a Comment